effect-uai

Effectful building blocks for agentic ai

[Jan @janschultecom]

a bit late to the party but @effectuai_sdk is now on @EffectTS_ v4.0 🥳

[Jan @janschultecom]

Coming soon to @effectuai_sdk
https://t.co/j2rr4HyVD8

[Jan @janschultecom]

TV stations coming to @effectuai_sdk! This has been quite fun: the AI generates a random TV programme on demand and then uses @MiniMax_AI h3-max-turbo on @fal to generate videos on the fly. Could potentially run endlessly until the credit card burns through 😅

[Jan @janschultecom]

Allrightyy. @effectuai_sdk now officially supports @typesafeai 's decision model jev!

Capability design heavily inspired by @tim_smart 's implementation for standard @EffectTS_ ai package, so kudos to him 👏

Alongside we ship a new recipe ticket triage to show you how to use it 👇

[Jan @janschultecom]

We've released v0.16 of @effectuai_sdk for @EffectTS_ with support for live models like speech-to-speech or omni models. Currently supports @OpenAI GPT realtime and @GoogleDeepMind Gemini Live models.

We also added two recipes to show you and your coding harness how to use it:
- Realtime voice agent
- Camera assistant

See 🧵👇

[Jan @janschultecom]

@GoogleDeepMind 's Gemini 3.8 live just came out and I took it for a spin with @effectuai_sdk for @EffectTS_ !
Latency is just stunning with full end-to-end streaming and support for tool calling.

[Jan @janschultecom]

Experimenting with live (aka realtime speech-to-speech) models for @effectuai_sdk and latency is just 🔥

@EffectTS_ 's Stream abstraction makes working with live models such a breeze.

[Jan @janschultecom]

We've release v0.15 of @effectuai_sdk for @EffectTS_:

- New Messenger capability to connect your effect agent to a Messenger
- Three new providers for the Messenger capabiklity: slack, discord and telegram
- A new recipe messenger agent to show you (or your agent) how to plumb it together

See 🧵👇

[Jan @janschultecom]

Where are we going to see @EffectTS_ based agents next?
@effectuai_sdk

[Jan @janschultecom]

We've published v0.14 of @effectuai_sdk for @EffectTS_
This release adds support for image generation and editing capability as well as provider support for @fal

Also two new recipes shipped to try it out:
- Storyboard
- Conversational image editing

[effect-uai @effectuai_sdk]

RT @janschultecom: Since @effectuai_sdk got covered in this week's @EffectTS_ office hours podcast by @maxwellbrown and @MattiaManzati, we decid…

[effect-uai @effectuai_sdk]

We finally added MCP support including for the latest stateless version. Check it out https://t.co/AKBGdisltN

[Jan @janschultecom]

Coming soon to @effectuai_sdk :
@aisdk ui compatibility. So you can keep your frontend aisdk code while getting all the @effectuai_sdk and @EffectTS_ magic in the backend:
* an explicit loop that you control
* streaming metrics (live tok/s)
* proper stream aborting
* typed error handling & model fallback

[Jan @janschultecom]

We've released v0.8 of @effectuai_sdk for @EffectTS_ :

* Support for websearch with three providers (@ExaDevelopers , @perplexity_ai and @tavilyai)
* A streamlined interface with better type-safety

As well as three amazing new recipes:
* AI radio station
* Grounded answer
* Deep research

https://t.co/IeotmiFfgZ

[Jan @janschultecom]

@effectuai_sdk v0.6 is out with some exciting new features:

• sandbox support to run agent code securely
• revamped primitives to make them easier to read
• pronunciation control and multi-speaker dialogue for speech synthesis

And two new recipes:
• sleeper-agent - wait for a long-running tool call
• run,fix,repeat - run code with automatic code fixing

https://t.co/rK4hML5oYe

[Jan @janschultecom]

cool new recipe and sandbox support coming soon to @effectuai_sdk for @EffectTS_

The recipe:
Run, fix, repeat - a self-correcting code interpreter using a code sandbox